Survey on Energy Efﬁciency in Ofﬁce and Residential ... · have shown that real-time feedback on power consumption leads to a reduction of energy consumption by up to 10%. 2.3

Survey on Energy Efficiency in Office andResidential Computing Environments

Andreas Berl1, Gergo Lovasz1, Hermann de Meer1 andThomas Zettler2

1Computer Networks and Computer Communications, University of Passau,94032 Passau, Germany; e-mail: [email protected] GmbH, Munich, Germany

Received 2 February 2012; Accepted: 10 April 2012

Abstract

Energy efficiency of computing equipment in office and residential environ-ments gets more and more important, with respect to the world-wide desireto reduce CO2 emissions and the increasing cost of energy. While hardwareitself gets cheaper, the cost of energy begins to dominate the total cost ofownership of a product. This paper gives an overview on energy savingmethods that are applied today, with a special focus on office and residentialenvironments. Currently used methods are classified into three categories:(1) autonomous management of devices, (2) coordinated management ofdevices, and (3) coordinated management of services. Various implement-ations of these methods in office and residential environments are describedand compared to each other. The comparison illustrates possible directions offuture research in the area of energy efficiency.

Keywords: energy efficiency, office environments, residential environ-ments.

Journal of Green Engineering, Vol. 2, 255–272.c© 2012 River Publishers. All rights reserved.

256 A. Berl et al.

1 Introduction

Energy-efficient Information and Communication Technology (ICT) isfostered by labels such as the U.S. Energy Star [8] or the European TCOCertification [16, 22, 35]. Furthermore, regulations as the European Energyrelated Products Directive [4] and the European Codes of Conduct [36] rateIT equipment according to their environmental impact.

The strict separation between the often used terms energy efficiency forICT (making ICT energy efficient) and ICT for energy efficiency (using ICTto achieve energy efficiency) [52] is vanishing in the area of residential andoffice environments. More and more non-IT equipment joins the network(i.e., gets an IP-address and becomes manageable [42, 44]), as the Internetof Things embraces more and more devices. This opens up the opportunityto save energy in classical non-IT equipment as well as in IT equipmentby using the same management mechanisms. The U.S. Energy InformationAgency [17] reports, that home electronics including IT equipment as PCsand entertainment TV sets account for 7% of the electricity consumed byU.S. households. Moreover, the European Eco-Design Directive [4] showsin recent studies [31] that IT equipment as PCs, peripherals, printers orphones exhibits in total consumes more energy than data centers. The car-bon footprint that is related to usage and directly corresponds to the energyconsumption is shown to be 259 Mt CO2 in 2002 and predicted to be 640 MtCO2 in 2020 (60 and 59% share of the global ICT footprint). For example theTelecom device’s global footprint was 18 Mt CO2 in 2002 and is expected toincrease almost threefold to 51 Mt CO2 by 2020 driven mainly by rises in theuse of broadband modems/routers and IPTV boxes.

Obviously the energy-saving potential in residential and office computingenvironments is huge, but due to their distributed nature and heterogeneousdevice landscape hard to exploit. This paper analyses energy-saving meth-ods that are available for IT equipment. It classifies current work into threemain categories of energy-saving methods: (1) Autonomous management ofdevices enables the reduction of energy consumption locally at a single device(e.g., by built-in energy-efficiency features). (2) Coordinated management ofdevices enables the optimization of the energy consumption of a group ofdevices that actively exchange energy-related information. (3) Coordinatedmanagement of services enables the replacement, delegation, and consolida-tion of services and aims at optimizing the energy consumption of a serviceor a group of services. In addition, this paper explores and compares imple-

Energy Efficiency in Office and Residential Computing Environments 257

mentations of energy-saving methods of each category in the context of officeand residential environments.

The remainder of this paper is structured as follows: Section 2 categor-izes current energy-saving methods. Sections 3 and 4 analyse the applic-ation of these methods in office and residential environments. Section 5provides a comparison of energy-saving methods concerning their use in bothenvironments, and Section 6 concludes this paper.

2 Energy-Saving Methods

This section identifies three disjunctive categories of energy-saving methodsthat reduce the energy consumption of devices. For each category severalexamples are described.

2.1 Autonomous Management of Devices

Autonomous management of devices covers energy management meth-ods that reduce the energy consumption of a device without coordinationwith other devices or the user. Instead, the device exploits its built-in en-ergy efficiency features autonomously. Dynamic external condition adaptionmonitors conditions that are caused externally (as CPU-workloads, CPU-temperatures, or user-interaction) and manages parts of a device accordingly.

The goal of the adaption is to dynamically adapt the managed device to itsenvironment in a way that the energy consumption of the device is reduced.It is important to see that this happens without a conscious interaction of theuser. Examples of dynamic external condition adaption are:

• A monitor is dimmed in reaction to low light conditions.• A fan is slowed down if the CPU is below a certain temperature.• Hardware parts are incrementally turned off due to sensing a lack of

user-machine interaction (e.g., display or disk).

2.2 Coordinated Management of Devices

In contrast to the autonomous management of devices, the coordinatedmanagement of devices addresses the cooperation between devices.

Automatic coordination reduces the energy consumption of a set ofdevices by exchanging energy-related information that eases up energy man-agement decisions. The purpose is to reduce the energy consumption of awhole set of devices instead of locally optimizing the energy consumption for

258 A. Berl et al.

each single device. Inter-device coordination can be achieved in a centralizedor decentralized way. In a centralized coordination approach, a centralizedentity either polls information from the managed devices or the manageddevices inform the managing entity periodically or at the occurrence of anenergy-relevant event. Based on the gathered information and policies, thecentral entity instructs devices to apply power saving methods. Coordinationcan also be achieved in a decentralized way, where energy-related informa-tion is exchanged, but decisions are made based on the local view of eachdevice. User-based coordination is triggered by implicit or explicit interac-tion between user and device. On one hand, the device may push informationto the user, e.g., a visualization of the current energy consumption of thedevice. On the other hand, the user is able to directly control the device,e.g., by sending the device to hibernation mode actively. Besides the en-ergy savings that are directly achieved by this approach, there are additionalpsychological effects that foster the energy-efficient behaviour of a user:Immediate feedback on the effects of his actions motivates energy-efficientbehaviour. Also competitive situations between users may be established,further motivating users to behave energy efficiently.

Examples of coordinated management of devices are:

• Cisco’s EnergyWise [33] (see Section 3.2) represents a centralized man-agement approach. A centralized server powers up/down groups devices,e.g., according to working/non-working times.

• The Energy Efficient Ethernet (EEE) standard (IEEE 802.3az [5]) is anexample of decentralized management approach. During times withoutdemand of data transmission, devices negotiate a low-power idle mode.

• Products as Kill-A-Watt [45] or Watts Up [19] (see Section 3.2) areproducts that support user-based coordination. They adapt their energyconsumption to user behaviour and visualize consumed energy.

• Projects that have been performed in residential environments [43, 51]have shown that real-time feedback on power consumption leads to areduction of energy consumption by up to 10%.

2.3 Coordinated Management of Services

Although services (e.g., print-servers, Open VPN servers, peer-to-peer cli-ents, or user desktops) do not consume energy directly, they utilise devicesand cause energy consumption indirectly. This category of methods re-duces the energy consumption of services, by replacing, delegating, orconsolidating them.


Service replacement is an approach for energy saving were services arereplaced by more energy-efficient services that provide the same (or similar)functionality. Although the energy-saving effect of service replacement canbe large, the overall impact on energy consumption is difficult to assess. Itmay even happen that the overall energy consumption increases, if the so-called rebound effect occurs. This effect describes the situation that a newenergy efficient service is so attractive to users that a high demand is createdwhich partially or fully compensates for the energy-saving effect of the re-placement. Service delegation allows the transfer of a service from one deviceto another, e.g., from a non-energy efficient to an energy-efficient device orto an always-on device (e.g., a router). The main goal of service delegationis to allow under-utilised devices to delegate their services to other devicesand change to an energy-saving mode. Service consolidation is based on theability of devices to process more than a single service at the same time.The goal of service consolidation is to reorganise the service to device map-ping within a group of devices in order to minimise the number of utiliseddevices. This means that the utilisation of some devices is increased whileother devices are relieved from their duties. Unutilised devices are hibernatedto save energy. Service consolidation can be done statically or dynamically. Ifit is done statically, a set of devices is determined that processes all requiredservices. If the external circumstances change, the allocation of the services isnot dynamically adapted. Dynamic service consolidation, in contrast, allowsfor the relocating of devices when external circumstances change, e.g., theloads of services change, or a device fails. Examples of service replacement,delegation, and consolidation are:

• Terminal servers [13,14,20] and virtual desktop infrastructures [3,18,21](see Section 3.3) replace user desktops in office environments. Insteadthe desktops are consolidated on servers within the data centre.

• Virtual private network server services can be delegated to the homegateway (router) in residential environments.

• Cloud computing achieves energy efficiency [28, 37, 48] (see Sec-tion 4.3) by consolidating user services (e.g., storage services) withindata centres. Cloud providers achieve a high utilisation of hardware andcustomers can dynamically allocate and release resources in the cloud.

260 A. Berl et al.

3 Office Environments

Office computing environments consist, e.g., of office hosts, network, peri-pheral devices as monitors, printers, scanners, and IP-phones. Within officeenvironments, especially office hosts contribute significantly to the IT relatedenergy consumption. On one hand, there is a high number of such hosts be-cause usually each employee typically has his own host. On the other hand,office hosts are often turned on 24/7. Webber et al. [50] have analyzed sixteenoffice sites in the U.S. and reported that 64% of all investigated office hostswere running during the nights. Although such hosts are mostly idle (CPUusage of 0%) during the time they are turned on, it is important to see thatthey still consume a considerable amount of energy. Measurements that havebeen performed at the University of Sheffield [32] show that typical officehosts which are idle still consume 49 to 78% of the energy that they needwhen they are intensely used, leading to an immense waste of energy.


Current office computing equipment often has the ability of saving energyby falling into low-power states if it remains unused for a critical period oftime. Hosts, monitors, or printers are dynamically hibernated to save energy.The low-power states of office hosts can be configured by the user and kickin when a host is idle for a critical time period. The Advanced Configurationand Power Interface (ACPI) specification [38] defines four different powerstates that an ACPI-compliant computer system (e.g., an office host) can bein. These states range from G0-Working to G3-Mechanical-Off. The statesG1 and G2 are subdivided into further sub-states that describe which com-ponents are switched off in the particular state. Separate power states (D0-D3for sub-devices and C0-C3 for CPUs) are defined, similar to the global powerstates [7, 34, 38]. However, as a matter of fact, many devices that are low-power capable do not successfully enter these states. Low-power modes aresubject to the complex combined effects of hardware, operating systems,drivers, applications – and after all – the user-based power management con-figuration. Webber et al. [50] report that in the investigated offices only 4%of all hosts actually have switched to low-power modes during the night.


In office environments power management solutions are able to optimisethe energy consumption of hosts that remain turned on while their users are


absent. Examples of this approach are eiPowerSaver [6], Adaptiva Compan-ion [2], FaronicsCore [9], KBOX [11], or LANrev [1]. office-wide powermanagement policies are applied in such approaches. Office hosts are forcedto adopt power management configurations, independent of user settings.Therefore, idle hosts can be set to a low-power state or be powered off to saveenergy. Additionally, often mechanisms are provided by such approaches towake up hosts if necessary (e.g., based on Wake-on-LAN technology). Thisway, inactive hosts can be accessed for administrative jobs (e.g., backupsthat happen during the night) and for remote usage. Cisco’s EnergyWisecontrols office equipment that is powered by Power over Ethernet (PoE). Itcan be used to power down IP-phones during nights and to power them upagain in the mornings. Additionally, EnergyWise can be used to apply energymanagement to hosts and to report energy savings within the office.

There are also user-based coordination methods available in offices: Theapproach of Greentrac [12] is setting its focus on the user’s energy awareness.A user is periodically informed about the energy consumption of the deviceshe is using. If the user is aware of the energy consumption he causes, he is ableto change his behaviour in order to save energy. The Greentrac-approach usesincentives to motivate the employees to implement energy-saving measures.


A typical example of service replacement and consolidation within officeenvironments, is the replacement of energy-consuming office hosts by highlyenergy-efficient thin clients [49]. Terminal-server approaches, e.g., move userdesktops to centralized terminal servers that are able to serve multiple userssimultaneously (consolidation). Terminal-server solutions are based on multi-user concepts where several users are able to log-on to a single OS that isprovided by the terminal server. OS, applications, and user data are stored inthe data centre and can be remotely accessed by thin clients. Common ter-minal server software products are Citrix XenApp [20], Microsoft WindowsServer 2008 [14], or the Linux Terminal Server Project [13].

In the Virtual Desktop Infrastructure (VDI) approach each user gets hisown Virtual Machine (VM). Similar to terminal servers, the VMs are storedwithin the data centre and can be accessed remotely by energy-efficient thinclients or any host with remote desktop software. In contrast to the terminalserver approach, the VDI approach has the advantage that each user can util-ise his preferred OS and individual applications (not all standard applicationsare able to run on terminal servers) and new virtualised desktops can be

262 A. Berl et al.

Figure 1 Common and virtualised office.

easily deployed. Furthermore, the virtualised desktops are strictly isolatedfrom each other, while being managed within the data centre. However, itis important to see that the provision desktop environments within VMs isdemanding: All of the VMs need a sufficient amount of CPU cycles, RAM,disc I/O, and other hardware resources to operate. Therefore, the number ofVMs that can be provided by a single server is rather limited. VDI productsare, e.g., VMWare View [3], Citrix XenDesktop [21], or Parallels VirtualDesktop Infrastructure [18].

In [24–26,30], a virtualised office environment is suggested that achievesa consolidation of services within office environments, independent of datacentre equipment. Office hosts are virtualised and virtual desktops are con-solidated dynamically on office hosts. Whereas terminal server and VDIsolutions impose changes to the office environment (thin clients replace fullfeatured office hosts), the virtualised office environment utilises availableoffice hosts. Terminal servers and VDIs, instead, move office services intothe data centre, which has two main disadvantages: First, additional hard-ware needs to be purchased and managed, second, the additional data centre


hardware consumes energy itself. Data centre equipment typically consumesmore energy than desktop hosts [41], due to high-performance parts, partsthat provide redundancy, and, especially, the cooling that needs to be appliedwithin the data centre.

In Figure 1 the transition from an ordinary to a virtualised office environ-ment is illustrated. It can be observed in the upper part of the figure that in theordinary office environment the Personal Desktop Environments (PDEs) andthe hosts are interdependent. Seven hosts are turned on together with sevenPDEs and three hosts (with PDEs) are turned off. The situation is differentin the virtualised office environment shown in the lower part of the figure.Although the number of currently running PDEs is the same as before, onlyfour hosts are actually turned on. It can be observed, e.g., that the upper righthost is providing three PDEs to users simultaneously.

Possible savings of about 50% of energy are reported with this ap-proach [24]. In comparison to VDI solutions which are able to save energyfor office environments with more than 25 hosts [40], the virtualised officeenvironments saves already energy in offices with only 4 hosts.

4 Residential Environments

More and more end-users have residential computing networks that consistof desktops, laptops, game pads, or home theater PCs. Such devices provideall kind of applications, including client, server, or peer-to-peer. Typically,residential networks have a gateway (residential gateway) to the Internet, e.g.,a Digital Subscriber Line (DSL) router.

Although office and residential environments seem to be similar on a firstglance, there are some major differences: in the office environment ratherhomogeneous office hosts are interconnected via Fast/Gigabit Ethernet andadministered by a professional administrator. In residential environments,highly heterogeneous hosts are usually connected to the Internet via DSL-connections, which typically have asynchronous up/download capacities andare administered by individual users. The access technology can be both,wired as well as wireless. In residential environments, the users may be chil-dren or adults with certain rights and restrictions in the home network. Thisprevents the use of uniform security policies.

264 A. Berl et al.


Similar to hardware of office environments also residential equipment per-forms a dynamic adaption to external conditions. Home IT-equipment (aslaptops or printers) is able to sense a lack of user interaction in order to turninto a hibernation mode. The residential gateway is a device that is constantlyturned on and typically provides a wide range of autonomous power man-agement features. Even in the case of no user interaction, a large numberof functions have to remain active to guarantee good user experience and tofulfill industry standards:

• A DSL-IP connection is required to receive VoIP calls. To transmit theIP stream the physical layer has to be kept active.

• The WLAN base station has to transmit beacon signals in order to per-form the association of new mobile devices to the WLAN network andto maintain the wireless link to previously associated devices.

• Ethernet link detection has to be active and attached devices have to bemanaged when requesting a new link.

• For the DECT/CatIQ cordless telephony interface the incoming calldetection has to be assured for all interfaces.

• The attachment of new devices to USB needs to be detected.

The list indicates some minimum active functions which have to be main-tained during autonomous management. In addition, other services may berequired to like for example FTP server functionality, multimedia server orhome automation functions. If the user actively decides not to use WLANduring night time,e.g, it can be turned off by using an autonomous timer basedmanagement that is configurable by the user.


Home automation, e.g., provides a coordinated management of non-ITdevices in the residential environment. Standards as G.hn [23] allow the con-nection of devices over any wire (power line, coax cable, phone line) or Wi-fiwireless connection.

Studies of the U.S. Energy Information Agency [17] show that majorenergy consumption in households stems from classical non-IT equipment.Therefore home automation opens a promising new opportunity for powersaving by including additional information for example from sensor networksto energy management decisions.


Figure 2 Home automation network.

An example of a home automation network architecture is illustrated inFigure 2, where a residential gateway is connecting multiple physical mediaforming together a heterogeneous network reaching virtually every control-lable device. A DSL or Passive Optical Network (PON) interface offers WideArea Network (WAN) services using broadband residential network. G.hnstandard ports are used for coaxial cable and powerline communication.Phones and phone-line devices are included via an inter-domain bridge. Eth-ernet LAN and USB devices are also covered. Residential gateways allowthe processing of automation applications inside the gateway independentfrom running PCs, which substantially saves power. Residential gatewaysare typically “always on” devices and form a natural central point for homenetworking.

Also user-based coordination is achieved within the residential environ-ment. Emerging technologies as, e.g., smart metering approaches raise theuser’s energy consumption awareness. Energy consumption can be monitoredlocally and remotely [47]. This is important to keep users aware about theirenergy behaviour. Having real-time feedback on current energy consumptionallows the user to link his actions to an increase of energy consumption.Taherian et al. [46], e.g., describe an energy monitoring system for resid-

266 A. Berl et al.

Figure 3 FHE architecture.

ential and office environments that supports continuous real-time feedbackon energy consumption.


Also service delegation and consolidation is achieved in residential envir-onments. In this field the residential gateway plays a key role. Residentialgateways mediate access to the Internet, run services on the user’s behalf, andcontrol IT and non-IT equipment. Furthermore, residential gateways are ableto support the coordinated management of services and are able to provideservices themselves energy efficiently: Residential gateways can provide ac-cess to peripheral devices as USB-Disks, printers, Network Attached Storage(NAS), multiport routing, switching and encryption, VoIP telephony func-tions (e.g., DECT/CatlQ or VoIP FXS/FXO H.323), or they can run P2Psoftware as BitTorrent or eDonkey.

Based on such residential gateways, Berl and co-authors [27–29, 39] de-scribe a Future Home Environment (FHE) that enables an energy-efficientconsolidation of services in home networks. It suggests the sharing of homenetwork resources with users of other home networks, similar to Grid com-puting approaches. Load is shifted to a small number of computers, in orderto relieve others. Unloaded computers are hibernated or turned off. A homenetwork is called active if it contains at least one computer which is turnedon and can share resources. In a passive home network only the gateway ison-line and other hardware is hibernated or turned off.

The FHE architecture is illustrated in Figure 3. In this example four homenetworks are interconnected by the FHE overlay, two active and two passivehomes. In the figure load is migrated from an end-host in the active home


network (b) to an end-host in the active home network (c). The end-host inhome network (b) can be hibernated or turned off after the migration process.If no further computer is turned on in home network (b), it can change itsstatus to passive.

5 Comparison

Most of the energy-saving methods that have been described in Section 2 areapplied in office and residential computing environments (see Sections 3 and4). Whereas the autonomous management of devices is implemented simil-arly in office and residential environments (hibernation of unused devices),the coordinated management of devices is implemented diversely in the twoenvironments. On the one hand, office-wide power management approachesand the controlling of PoE devices is applied in office environments. Suchmechanisms can be easily applied, due to the rather homogeneous officecomputing environment and similar usage patters of office users, which easesup the energy management: Sets of devices can be hibernated, e.g., ac-cording to time-based energy-saving policies (working/non working times).Multipurpose devices of residential networks, on the other hand, are ratherheterogeneous and the behaviour of the users is less predictable. The man-agement of the devices needs to be done in a context-aware way, where thebehaviour of the users is monitored in order to take management decisions.Also waking up devices is often easier in office environments as Wake-on-LAN can be applied to devices that are connected to wired networks. Inresidential environments many devices are attached wireless to the homegateway which makes it hard to wake them up remotely. Instead, home auto-mation systems are applied that mainly adapt energy consumption of non-ITdevices to the behaviour of persons in a household.

The user-device interaction for energy-saving requires users sufficientlytrained and with awareness of energy consumption. The typical residential-user will need a simple and easy to use interface that provides direct feedbackon energy consumption. Independent of residential or office environment thegeneral consciousness of energy saving is key to motivate user to take action.

The application of coordinated management of services, however, is un-balanced between office and residential computing environments: Servicedelegation is typically applied in residential environments. Home Gatewaystake over services as, e.g., printing servers, network storage servers, or serversfor DECT phones and needs to be further exploited in office environments.Together with the novel approach of the Virtualized Office Environment (as

268 A. Berl et al.

described in Section 3.3) the energy consumption of offices can be reduced,especially if no data centre infrastructure is available. Whereas service re-placement and consolidation has already been applied widely in the area ofoffice environments (in terms of virtual desktop infrastructures and terminalservers), it is not yet sufficiently exploited in residential environments. Novelapproaches, as the Future Home Environment (see Section 4.3), provide ahigh potential of energy saving in this area. Also the paradigm of cloud com-puting can be further exploited: Instead of thin-clients, users of residentialnetworks are able to use energy-efficient equipment as smart-phones, tablets,or netbooks to access cloud-based services. Gaming PCs, desktops, or hometheater PCs may not be needed anymore in future residential network scen-arios. There are already some approaches available, as e.g. OnLive [15] orGAIKAI [10], where demanding 3D games can be played within the cloud.

6 Conclusion

This paper has reviewed the state of the art of available energy-saving meth-ods, especially concerning office and residential environments. Currentlyapplied methods have been categorized into three classes: (1) The autonom-ous management of devices allow devices to reduce their energy consumptionindividually, without active cooperation of other devices or humans. (2) Co-ordinated management of devices covers automatic management as wellas user interaction. In contrast to the first category, such strategies saveenergy through cooperation, using energy-relevant signalling to exchangeinformation. Groups of devices are managed cooperatively (or are managingthemselves) to achieve the common goal of a reduced energy consumption.Finally, (3) the coordinated management of services performs the replace-ment, delegation, and consolidation of services in a cooperative way to reducethe overall energy consumption.

Although all of the mentioned categories are applied within office envir-onments as well as in residential environments, their application is unbal-anced. Especially, the coordinated management of services provides a highpotential of energy savings that can be exploited by future developments.

Acknowledgements

The research leading to these results has been partly supported by the GermanFederal Government BMBF in the context of the G-Lab Ener-G project, by


the ECs FP7 All4green project (grant agreement no. 288674), by the FP7EuroNF Network of Excellence (grant agreement no. 216366, Joint SpecificResearch Project EEWMI) and by the COST Action IC0804.

References[1] Absolute MANAGE. http://www.lanrev.com/solutions/power management.html.[2] Adaptiva Companion. http://www.adaptiva.com/products companion.html.[3] Desktop Virtualization: VMware View. http://www.vmware.com/products/view.[4] Directive 2009/125/EC. http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:

2009:285:0010:0035:en:PDF.[5] Draft Amendment to IEEE Std 802.3-2008,IEEE Draft P802.3az/D2.0, IEEE

802.3az Energy Efficient Ethernet Task Force. http://grouper.ieee.org/groups/802/3/az/public/index.html.

[6] eiPower Saver Solution. http://entisp.com/pages/eiPowerSaver.php.[7] Energy Star for External Power Adapters. http://www.energystar.gov/ia/partners/

product specs/program reqs/eps prog req.pdf.[8] EU Energy Star: European Commission Directorate-General for Energy. http://www.eu-

energystar.org.[9] FARONICS: Intelligent Solutions for ABSOLUTE Control.

http://faronics.com/html/CoreConsole.asp.[10] GAIKAI Open Cloud Gaming Platform. http://www.gaikai.com.[11] Green IT. http://www.kace.com/solutions/green-it.php.[12] Greentrac – A new way to reduce the power your company uses by leveraging your

greatest resource: Your people. http://www.greentrac.com/index.php.[13] LTSP: LINUX Terminal Server Project. http://www.ltsp.org.[14] Microsoft Windows Server 2008 R2. http://www.microsoft.com/windowsserver2008.[15] OnLive desktop. http://www.onlive.com.[16] TCO Certification. http://www.tcodevelopment.com.[17] U.S. Energy Information Agency, Household Electricity Report, U.S. Department of

Energy. http://www.eia.doe.gov/emeu/reps/enduse/er01 us tab1.html.[18] Virtuelle Desktop-Infrastruktur. http://www.parallels.com/solutions/vdi.[19] watts up? https://wattsupmeters.com.[20] XenApp. http://www.citrix.com/XenApp.[21] XenDesktop 5: The virtual desktop revolution is here . . . for everyone.

http://www.citrix.com/virtualization/desktop/xendesktop.html.[22] Preparatory studies for Eco-design Requirements of EuPs Lot 3 Personal Com-

puters (desktops and laptops) and Computer Monitors. http://www.ebpg.bam.de/de/ebpg medien/003 studyf 07-08 complete.pdf, 2007.

[23] International Telecommunication Union, New ITU standard opens doors for uni-fied ‘smart home’ network. http://www.itu.int/newsroom/press releases/2009/46.html,October 2009.

[24] Andreas Berl. Energy Efficiency in Office Computing Environments. PhD Thesis,University of Passau, Fakultat fur Informatik und Mathematik, May 2011.

270 A. Berl et al.

[25] Andreas Berl and Hermann De Meer. A virtualized energy-efficient office environment.In Proceedings of the ACM SIGCOMM 1st Int’l Conf. On Energy-Efficient Computingand Networking (e-Energy 2010), pages 11–20. ACM, April 2010.

[26] Andreas Berl and Hermann De Meer. An energy-consumption model for energy-efficientoffice environments. FGCS, 27(8):1047–1055, October 2011.

[27] Andreas Berl, Hermann De Meer, Helmut Hlavacs, and Thomas Treutner. Virtualiza-tion in Energy-Efficient Future Home Environments. IEEE Communications Magazine,47(12):62–67, December 2009.

[28] Andreas Berl, Erol Gelenbe, Marco Di Girolamo, Giovanni Giuliani, Hermann De Meer,Minh Quan Dang, and Kostas Pentikousis. Energy-efficient cloud computing. TheComputer Journal, 1–7, August 2009.

[29] Andreas Berl, Helmut Hlavacs, Roman Weidlich, Michael Schrank, and HermannDe Meer. Network virtualization in future home environments. In Proc. of the 20thInt’l Workshop on Distributed Systems: Operations and Management (DSOM09), Lec-ture Notes in Computer Science (LNCS), Vol. 5841, pages 177–190. Springer Verlag,October 2009.

[30] Andreas Berl, Nicholas Race, Johnathan Ishmael, and Hermann De Meer. Networkvirtualization in energy-efficient office environments. Computer Networks, 54(16):2856–2868, November 2010.

[31] Mic Bowman, Saumya K. Debray, and Larry L. Peterson. SMART 2020: Enablingthe low carbon economy in the information age. ACM Trans. Program. Lang. Syst.,15(5):795–825, November 1993.

[32] Chris Cartledge. Sheffield ICT Footprint Commentary. http://www.susteit.org.uk/files/files/26-Sheffield ICT Footprint Commentary Final 8.doc, 2008.

[33] CISCO. Cisco EnergyWise Technology. http://www.tcodevelopment.com.[34] European Commission. EU Directive for external power supplies. Official Journal of the

European Union, April 2009.[35] European Commission – DG TREN. CONNECT: Coordination and Stiumulation of

innovative ITS activities in Central and Eastern European countries. http://www.connect-project.org/index.php?id=35.

[36] European Commission, Joint Research Centre. Code of conduct on energy consumptionof broadband equipment, February 2011.

[37] Carl Hewitt. Orgs for scalable, robust, privacy-friendly client cloud computing. IEEEInternet Computing, 12(5):96–99, 2008.

[38] Hewlett-Packard. Microsoft, Phoenix, and Toshiba. Advanced configuration and powerinterface specification. ACPI Specification Document, 3, 2004.

[39] Helmut Hlavacs, Karin A. Hummel, Roman Weidlich, Amine M. Houyou, Andreas Berl,and Hermann De Meer. Distributed energy efficiency in future home environments.Annals of Telecommunication: Next Generation Network and Service Management,63(9):473–485, October 2008.

[40] IGEL Technology. Große IT-Strategie fur kleine Betriebe: VDI und Thin Clients. Pressrelease, 30.3.2010.

[41] J.G. Koomey. Estimating total power consumption by servers in the US and the world,February 2007.

[42] Jorg Luther. Living and Working in a Global Network. Springer, Berlin/Heidelberg, 2005.


[43] L.T. McCalley and C.J.H. Midden. Energy conservation through product-integrated feed-back: The roles of goal-setting and social orientation. Journal of Economic Psychology,23(5):589–603, 2002.

[44] Steve Meloan. Toward a Global Internet of Things. Technical Report, Sun DeveloperNetwork, November 2003.

[45] P3 International. Welcome to P3 International. http://www.p3international.com.[46] G. Coulouris, S. Taherian, M. Pias, and J. Crowcroft. Profiling energy use in households

and office spaces. In Proceedings of 1st International Conference on Energy-EfficientComputing and Networking 2010, pages 21–30. ACM, New York, USA, 2010.

[47] H. Siderius and A. Dijkstra. Smart metering for households: Costs and benefits for theNetherlands. In Energy Efficiency in Domestic Appliances and Lighting, p. 207, 2006.

[48] L.M. Vaquero, L. Rodero-Merino, J. Caceres, and M. Lindner. A break in the clouds: To-wards a cloud definition. ACM SIGCOMM Computer Communication Review, 39(1):50–55, 2008.

[49] Willem Vereecken, Lien Deboosere, Pieter Simoens, Brecht Vermeulen, Didier Colle,Chris Develder, Mario Pickavet, Bart Dhoedt, and Piet Demeester. Power efficiency ofthin clients. European Transactions on Telecommunications, 13:1–13, 2009.

[50] C.A. Webber, J.A. Roberson, M.C. McWhinney, R.E. Brown, M.J. Pinckard, andJ.F. Busch. After-hours power status of office equipment in the USA. Energy – TheInternational Journal, 31(14):2487–2502, 2006.

[51] G. Wood and M. Newborough. Dynamic energy-consumption indicators for domesticappliances. Energy and Buildings, 35(8):821–841, 2003.

[52] Mazin Yousif. Towards green ICT. In ERCIM News online edition. European ResearchConsortium for Informatics and Mathematics (RCIM), October 2009.

Biographies

Andreas Berl obtained his Ph.D. at the University of Passau (Germany)in 2011. He is currently working as researcher in the Computer Networksand Communications group at the University of Passau, chaired byProfessor Hermann de Meer. His research interests include energy efficiency,virtualization, and peer-to-peer overlays. Currently he is involved in theBMBF project “G-Lab Ener-G – Improving the Sustainability of G-Labthrough Increased Energy Efficiency” and in the EU projects “FIT4Green– Federated IT for a sustainable environmental impact” and “All4Green –Active collaboration in data centre ecosystem to reduce energy consumptionand GHG emissions (FP7)”. Andreas Berl is member of the EU Network ofExcellence “EuroNGI/EuroFGI/EuroNF – Design and Engineering of theNext Generation Internet” and the COST Action IC0804 “Energy Efficiencyin Large Scale Distributed Systems”. In 2009 he had a DAAD scholarship atLancaster University, UK, supervised by Professor David Hutchison.

272 A. Berl et al.

Gergo Lovasz received his master degree in computer science in 2008 atthe University of Passau (Germany). Currently, he is Ph.D. student at theChair of Computer Networks and Communications headed by ProfessorHermann de Meer at the University of Passau. His main research area isenergy efficiency in large-scale distributed systems. Currently he is workingon the research project “G-Lab Ener-G”, funded by the German FederalMinistry of Education and Research (BMBF). He is member of the EuropeanNetwork of Excellence EuroNF and the COST Action IC0804 “EnergyEfficiency in Large Scale Distributed Systems”. In 2010 and 2011 he waslocal organization chair of the e-Energy conference series on energy-efficientcomputing and networking. At e-Energy 2012 he was member of the TPC.

Hermann de Meer is currently appointed as Full Professor of computerscience (Chair of Computer Networks and Communications) at theUniversity of Passau, Germany. He is director of the Institute of IT Securityand Security Law (ISL) at the University of Passau. He had been an AssistantProfessor at Hamburg University, Germany, a Visiting Professor at ColumbiaUniversity in New York City, USA, Visiting Professor at Karlstad University,Sweden, a Reader at University College London, UK, and a research fellowof Deutsche Forschungsgemeinschaft (DFG). He chaired one of the primeevents in the area of Quality of Service in the Internet, the 13th internationalworkshop on quality of service (IWQoS 2005, Passau). He has also chairedthe first international workshop on self-organizing systems (IWSOS 2006,Passau) and the first international conference on energy-efficient computingand networking (e-Energy 2010, Passau).

Thomas Zettler is Principal System Engineer, responsible for energy-efficiency and power management system-on-chip concepts at LantiqDeutschland GmbH. He is representative at the European Commission’s“European Code of Conduct for Broadband Equipment” and at the U.S. De-partment of Energy and U.S. Environmental Protection Agency ENERGYSTAR Small Network Equipment program. Before this he was Principal invarious leading concept and development positions at Infineon TechnologiesAG and in technology process development at Siemens AG. He holds a Dr.rer. nat. degree and Diploma degree in physics from the University of Ham-burg, Germany. He has authored and co-authored numerous publications andis a member of the European Design Automation Association.

Documents

Survey on Energy Efﬁciency in Ofﬁce and Residential ... · have shown that real-time feedback on power consumption leads to a reduction of energy consumption by up to 10%. 2.3